The article compares the effect of selective and universal screening on detection rate and outcomes of pregnancies complicated by diabetes mellitus (DM) in a multiethnic population. The method used was to review the pregnancy and delivery of two 18-month periods, 5 years apart. In the year 1996-1997 when selective screening was used 315 (5.7%) of 5506 delivered women had diabetes during pregnancy. The rates of diabetes in the different ethnic groups were: UAE (4.4%), Peninsula Arabs (4.0%), Chami Arabs (4.5%), North African Arabs (6.7%), Indian subcontinent (7.5%), and Somalis and Sudanese (9.7%). The rate of diabetes among the different ethnic groups for the year 2001-2002 when screening was universal but diagnosis made by the same criteria were 590 (9.7%) of 6232 delivered women, UAE (9.2%), Peninsula Arabs (8.4%), Chami Arabs (8.2%), North African Arabs (9.6%), Indian Subcontinent (11.0%), Somalis and Sudanese (11.3%). The outcome indicators and their rates in the years 1996-1997 and 2001-2002 were respectively: gestational diabetes, 86.3%, 89.0%; requirement of insulin treatment, 74.3% 82.5%; vaginal delivery, 68.2%, 75.3%; cesarean section, 30.3%, 19.8%; macrosomia, 22.2%, 6.7%; intrauterine fetal death, 2.9%, 1.1%; and preterm delivery, 22.5%, 17.5%. This article confirms the influence of ethnic background on the prevalence of gestational diabetes in a multiethnic and multicultural society. Over a period of 5 years, there was a 66.7% increase in the incidence of gestational diabetes, which was probably due to a combination of increased detection by change in screening policy and an increase in the incidence of gestational diabetes. The indicators of disease severity and control, such as insulin requirement, rates of abdominal delivery, macrosomia, and structural congenital malformations, were significantly better in a cohort identified by universal screening compared with that identified by selective screening. Universal screening seems to be a more appropriate strategy for screening in this environment.